On the size of the intersection of two Lucas sequences of distinct type. II (Q763640)

A Lucas sequence is a sequence of general term \(u_n=(\alpha^n-\beta^n)/(\alpha-\beta)\), where \(\alpha\) and \(\beta\) are such that \(\alpha+\beta=r\) and \(\alpha\beta=s\) are coprime integers. Obvious examples are the repunits \((b^n-1)/(b-1)\) in an integer base \(b>1\), for which \((\alpha,\beta)=(b,1)\), as well as the sequence of \(Y\)--coordinates in the positive integer solutions \((X,Y)\) of the Pell equation \(X^2-dY^2=1\), for which \((\alpha,\beta)\) are obtained from suitable powers of the fundamental unit and its inverse in the real quadratic field \(\mathbb Q(\sqrt{d})\). In the paper under review, the authors study when two such sequences can intersect. More precisely, they look at the positive integer solutions \((x,n)\) of the Diophantine equation \(x^2-a((b^n-1)/(b-1))^2=1\). They show that for fixed integers \(a>1\) and \(b>1\), the above equation has at most two positive integer solutions \((x,n)\). This improves a recent result from \textit{B. He, A. Togbé} and \textit{P. G. Walsh} [``On the size of intersection of two Lucas sequences of distinct type'', Ann. Sci. Math. Qué. 35, No. 1, 31--61 (2011; Zbl 1271.11037)], where it was shown that the same equation has at most three positive integer solutions. They also show that the number two can be further reduced to one in some cases like when \(b\) is odd, or when \(b\) is an even perfect square. The proofs are intricate using both a lower bound for a linear form in three logarithms due to Matveev, and a linear form in two logarithms due to Laurent, reduction techniques á la Baker and Davenport, as well as several hours of computer computations.